Validation method of non-probabilistic Bayesian model for aircraft structure

Abstract

The invention discloses a validation method of a non-probabilistic Bayesian model for an aircraft structure. The validation method comprises the following steps: firstly deducing a non-probabilistic Bayesian theory according to a proposed interval inclusion degree theory; then updating and validating an uncertainty numerical simulation model based on the proposed non-probabilistic Bayesian theory and in combination with validated test data and an uncertainty interval propagation algorithm. The validation method disclosed by the invention is performed under a Bayesian probability theory framework, prior information can be fully utilized, and the reliability of the numerical simulation model can be given; the model validation can be performed aiming at the situations of poor information and less data, so that the problem that a large amount of experimental data is needed when a traditional model confirmation method is utilized is solved.

Description

technical field [0001] The invention relates to the technical field of aircraft structure numerical simulation, in particular to a method for confirming a non-probability Bayesian model of an aircraft structure. Background technique [0002] The three major laboratories (LANL, SNL, and LLNL) affiliated to the US Department of Energy, Los Alamos, Sandia, and Laurels Livermore, introduced model verification and validation (V&V for short) into the famous model around 1998. The Accelerated Strategic Computing Initiative (ASCI) program, later the Advanced Simulation and Computing (ASC) program. In 2006, Lloyds Livermore Laboratory published a white paper on model validation and validation in the ASC program. In this white paper, the key to the validation and validation of numerical simulation models, the quantification of uncertainties, and the existence of model validation and validation Problems and future prospects are described in detail. The American Institute of Aeronauti...

AI Technical Summary

Problems solved by technology

For complex large-scale systems such as aircraft, it is usually impossible to provide enough test data to complete the confirmation of the whole system model

At present, there are many difficulties in research in the field of aircraft structure numerical simulation model confirmation. For example, in the process of aircraft structure design, manufacture and use, there are related uncertain factors such as material properties, geometric characteristics, processing technology, load conditions, etc. , and these uncertainties are difficult to quantify

However, the acquisition of uncertainties and numerical calculation errors is the weakest link in the numerical simulation of large complex structures

When using experimental data for model validation research, classical probability theory and statistics are the most commonly used methods, but this method requires a large sample size

For situations with poor information and little data, such as aircraft structures, bridges and large building structures, statistical methods are difficult to apply

Images